One way would be to put a unit pressure load on the shell and drive it with a time curve that is the sum of the dead load plus the live load.
What is the non-linearity in your problem?
Thanks for your helpful reply.
I implemented a very similar solution. I applied the constant pressure by using a slowly (10*system specific time) increasing linear curve and only then I added the pressure pulse. Thus the pressure pulse was delayed some tenth of seconds until the constant head was slowly increased to its constant value. I believe this is very similar to your suggestion.
Regarding the non-linearity, it is the material. Because I load the vessel above the yield stress into the plastic zone I am using a bi linear model of the material.
Any comment you may have about this solution is very welcomed.